1. Field of the Invention
The present invention relates to an image joining apparatus, an image joining method, and an image joining program. In particular, the present invention relates to an image joining apparatus, an image joining method, and an image joining program for creating a joined image by joining a plurality of consecutive still images on a time axis, the still images constituting a moving image.
2. Description of Related Art
A technique for forming a larger image by joining a plurality of still images or by joining still images constituting a moving image is known as photomosaicing or the like.
A technique for photographing a moving image by manually moving a camera to photograph a large image and for creating a joined image by joining still images constituting the moving image has been recently achieved. Such a technique is introduced in digital cameras and the like as an additional function. In particular, the use of software for automatically or semi-automatically performing joining processing has become widespread from the time that use of digital cameras became widespread.
For example, Japanese Unexamined Patent Application Publication No. 2013-30875 discloses a technique for joining still images taken along a free camera trajectory. According to this technique, a camera moving direction during photographing is determined and still images are joined together based on the camera moving direction, thereby generating a joined image.
However, the joining of still images taken along a loop-like camera trajectory has a problem that misregistration occurs in images that should be located at the same position in the still images located at a loop start point and a loop end point. This problem occurs due to the camera trajectory and accumulation of errors in motion vectors obtained from the photographed still images.
The accumulation of errors can be reduced by using a combination of pieces of information from, for example, a high-precision direction sensor, a posture sensor, and/or an acceleration sensor. However, it is difficult to incorporate a small sensor with a sufficiently high performance into a small, lightweight, and/or inexpensive device.
Although the accumulation of errors can be reduced by repeated processing involving correction of the positions of images, the amount of calculation for repeated processing is extremely large. Accordingly, it is difficult to incorporate a circuit for achieving the repeated processing into a small, lightweight, and/or inexpensive device.